A Kind of Optimization Method for Committee and Reward Mechanism

Consensus protocols play a vital role in blockchain by determining block generation speed and ensuring blocks’ consistency and security. To address issues such as low efficiency and inadequate throughput in consensus algorithms, we have conducted research and proposed a novel Byzantine Fault Tolerant (BFT) protocol called CF-BFT, which operates in a dual-mode fashion and relies on node identity authentication. The CF-BFT protocol encompasses two sub-protocols: CheckBFT and FastBFT. Initially, the system assumes a pessimistic environment and executes the CheckBFT protocol, wherein a committee verifies the current environment’s safety and the integrity of the leader node. Upon confirming the safety of the current environment, the CF-BFT protocol transitions to the FastBFT protocol for message handling. By optimizing the number of participating nodes, we optimized and reduced the participant nodes from the conventional \(3f+1\) to a more efficient \(2f+1\) committee members. This mitigates the communication overhead. By incorporating a hash-based selection mechanism for choosing the primary, we have enhanced the decentralization and randomness of the protocol. Moreover, the adoption of a ring topology for selecting primary nodes and committee members greatly enhances the scalability of the blockchain network. Furthermore, we introduce the concepts of cumulative reputation value and consumable reputation value to optimize the reward mechanism, resulting in a more rational tenure of the primary node within the FastBFT sub-protocol. Experimental results demonstrate that the proposed algorithm outperforms traditional BFT algorithms and other dual-mode algorithms across various performance indicators. While ensuring consistency and security, it significantly reduces communication costs and improves consensus efficiency. The experiment showed that in the best-case scenario, we improved by 47% to 2.5 times compared to SAZyzz, while in the worst-case environment, we achieved a maximum performance increase of 20% to 70%.